University of Arkansas, Fayetteville University of Arkansas, Fayetteville

ScholarWorks@UARK ScholarWorks@UARK

Mechanical Engineering Undergraduate Honors Theses Mechanical Engineering

5-2019

Silicone Tadpole: Research into Soft Bodies Silicone Tadpole: Research into Soft Bodies

Danielle Fernandez

Follow this and additional works at: https://scholarworks.uark.edu/meeguht

Part of the Applied Mechanics Commons, Computer-Aided Engineering and Design Commons,

Polymer and Organic Materials Commons, and the Service Learning Commons

Citation Citation Fernandez, D. (2019). Silicone Tadpole: Research into Soft Bodies. Mechanical Engineering Undergraduate Honors Theses Retrieved from https://scholarworks.uark.edu/meeguht/83

This Thesis is brought to you for free and open access by the Mechanical Engineering at ScholarWorks@UARK. It has been accepted for inclusion in Mechanical Engineering Undergraduate Honors Theses by an authorized administrator of ScholarWorks@UARK. For more information, please contact scholar@uark.edu.

Silicone Tadpole:

Research into Soft Bodies

An Honors Thesis Report

for the University of Arkansas

Department of Mechanical Engineering

By:

Danielle Fernandez

April 26, 2019

University of Arkansas

Presented to:

Dr. Steve Tung, Thesis Advisor

Department of Mechanical Engineering

University of Arkansas

Dr. James Davis

Department of Mechanical Engineering

University of Arkansas

1 | P a g e

Acknowledgements

I would first like to thank Dr. Tung for advising me through this process. With his help, I

was able to get thought-provoking results and data. This thesis would not have been possible

without his guidance and professional advice. Also, I would like to thank Dr. Davis for taking

time to listen to the defense of this thesis.

Finally, a special thanks to the Engineering Research Center and the Mechanical

Engineering Department at the University of Arkansas for providing essential tools and programs

that ensured the project’s completion. The silicone printer, Solidworks, ReplicatorG, and other

programs are examples of the tools that went into this project.

2 | P a g e

Table of Contents

Acknowledgements……………………………………………………………………………...1

List of Tables………………………………………………………………………………….....3

List of Figures…………………………………………………………………………………...4

List of Symbols………………………………………………………………………………….5

Abstract……………………………………………………….………………………………....6

I.Introduction…………………………………………………………………….………………..7

II.Background Research……………………………………………………………………….…...8

III.Goal and Objectives……………………………………………………...……………………..10

IV.Design Process…………………………………………………………...……………...……...11

A. Printer Specifications………………………….………………..11

B. Proposed Designs………………………………...……………..14

C. Analysis of Final Design..……………………………………....20

D. Testing of Final Design………………………………..……..…21

V.Test Results and Discussion……………………………………………………..……………...22

VI.Conclusion………………………………………………………………...……...……………..24

VII.References…………………………………………………………………………………..…..25

Appendix: Assembly Drawing(s), Exploded View, Component Drawings…….…………….27

3 | P a g e

List of Tables

Table 1: Theoretical Values for Theta based on Final Design…………………………………..21

Table 2: Experimental Data for Theta based on Specific Pressure………………………..…….22

Table 3: Percent Error between Experimental and Theoretical Values………………………….23

4 | P a g e

List of Figures

Figure 1: Cross-section of Tail with Variables……………………………………………………9

Figure 2: Silicone Backbone……………………………………………………………………..10

Figures 3 & 4: Physical Set-Up for a Wanhao Printer with Connections to Nitrogen Tanks……12

Figure 5: Print Settings…………………………………………………………………………..13

Figure 6: Printer and Filament Settings………………………………………………………….13

Figure 7: Tadpole Head Design………………………………………………...………………..15

Figure 8: First Design of Tail…………………………………………………………………….15

Figure 9: Printed Version of First Tail Design……………………………………………..……16

Figure 10: Second Design of Tail………………………………………………………………..17

Figure 11 and 12: Printed Version of Second Tail Design………………………………………18

Figure 13: Third Design of Tail………………………………………………………………….18

Figure 14 and 15: Printed Version of Third Tail Design………………………………………...19

Figure 16: Final Design of Tail…………………………………………………………………..20

Figure 17 and 18: Printed Version of Final Tail Design…………………………………………20

Figure 19: Data taken at 15 psi with a 10 degrees Angle of Deformation…………….…………22

Figure 20: Graph of Theorical and Experimental Angle of Deformation……………………….23

Figure 21: Full SolidWorks Design of Tadpole Head………………………………...…………27

Figure 22: Drawing of Tadpole Head……………………………………………………………27

Figure 23: Full SolidWorks Design of First Tadpole Tail……………………………………….28

Figure 24: Drawing of Final Design of Tadpole Tail……………………………………………28

Figure 25: Placement of Cross-Section in ReplicatorG………………………………………….29

Figure 26: Example of Problems with Silicone Printer i.e. Channel Collapse…………………..29

5 | P a g e

List of Symbols

ε Material Strain

σ Material Stress

θ Angle of Deformation

E Young’s Modulus

Psi Pound per square inch

n Number of Channels

w Width of Channels

h Height

c Channel

a Actuator

P Pressure

GPa Giga-Pascal

C Celsius

Si Silicon

O Oxygen

mm Millimeter

6 | P a g e

Abstract

In this thesis, research is conducted in the area of soft robotics by building a soft tadpole

that can deform with a specific air pressure. The goal is to mimic the motion of an organic

tadpole in respect to its S-shaped tail movement. The angle of deformation, derived from

material mechanic theories, ranges from 45 to 80 degrees for this type of movement. The design

includes a head compartment which acts as a tank to transfer nitrogen pressure and a tail section

that receives the said pressure and bends as a result. The tail section was designed with two rows

of air channels within a rectangular cross-section. Nitrogen that is pumped in the top row will

cause the empty bottom row to deform and create movement. The material assigned to the tail

section is silicone due to its flexible and water repellant properties. In order to print the tail, a

Wanhao printer was transformed into a 3D printer that produces silicone in its uncured state.

After many attempts to create a tail design that bends under pressure, a final rectangular cross-

section with 8 channels in each row was achieved. However, when this design was tested, the

angles of deformation did not match the theoretical values. This is due to a great amount of error

in the design and printing stage of the process. In the end, the tadpole was able to bend into a C-

shape formation and somewhat imitate the slow swimming movement of its organic counterpart.

7 | P a g e

I. Introduction

The industry of robotics has come far throughout the years, but the soft robotics sector

has taken a recent interest in the scientific community. Soft robotics deals with problems and

situations that rigid bodies simply cannot. This could mean many possibilities for the

technological market and for future innovations. Imagine a robotic device that could move

through fluid easily and efficiently in order to reach its goal. Or one that could maneuver through

crevices by the fluidity of its body. Either way, this trend in robotics allows for numerous

opportunities and applications for advancement.

This project is a starting point for this University to create their own soft robot. As it

happens, due to time constraint, the body of the robot and the material fluidity was the main

focus point rather than the robotics aspect. The robotics aspect will be continued by fellow

undergraduates and will use this design as a starting point. The robotic subject chosen to emulate

organic movements was a tadpole. Tadpoles have the perfect tail to test fluidity since it curves to

make an S-shape and not a C-shape like most fish. Also, they have a decently sized bulbous head

that can theoretically store the robotic components. Yet for this case, nitrogen pressure will be

manually pumped into the head apparatus for testing.

In order to ensure that the tail of the tadpole will deform and move, a silicone material

was chosen to make up most of the body. Silicone is a very ductile material with a number of

real-world applications. Additionally, this material was carefully picked due to its water

repellency and ability to withstand very cold temperatures. Since the tadpole will be testing in

water, these characteristics will prove to be an asset in the overall design. Through theories given

in material science, the tadpole tail is predicted to deform in a manner that copies the S-shape of

an organic tadpole.

8 | P a g e

II. Background Research

The inspiration for this project came from a scholarly paper written by students at MIT.

Their research proved that a self-contained autonomous robotic fish was possible to construct

and able to deform under fluid pressure. From their research, the best way for deformation in a

tail to occur would be through a series of parallel air channels. These channels will line the top

and bottom half of the tadpole tail. In order to simplify the model, only a cross-section of the tail

will contain the channels. The idea is to pump pressure or fluid into the top half and leave the

bottom half empty. This will create axial tension as the channels continue to expand, and

eventually after constant expansion, the lateral stress transforms into a bending moment. The rest

of the tail will be made out of pure silicone, but will follow the deformation pattern started by the

air channels. In reference to the theory provided by Marchese and colleges in their paper, the

total bending angle of a rectangular cross-section uses both the physical properties of the

channels and the external pressure. [1]

Θ = 2𝑛 𝑡𝑎𝑛−1(𝑤𝜀(𝜎)

2ℎ𝑐) (1)

σ =𝐹

𝐴= 𝑃𝑎(

ℎ𝑐

ℎ𝑎−ℎ𝑐) (2)

ε =𝜎

𝐸 (3)

These equations are based on stiff body mechanics. [4] In the first equation, the theta is

the angle of deformation that will occur when a pressure is applied throughout the system. The

variable n represents the number of channels, w is the channel width, ℎ𝑎 is the actuator height,

and ℎ𝑐 is the channel height. The height variables remain constant. Yet, for this project the

actuator height will be equivalent to the height of the head component. The second equation

calculates the material stress on the tail which will be used in the first equation as well. Finally,

9 | P a g e

the third equation gives the non-linear material strain which depends on the second equation and

Young’s modulus. Young’s modulus for silicone ranges from 0.001-0.05 GPa. In order to get

more accurate theoretical values, the average of the range was taken and that value was used in

calculations. All the variables and how they related to the tail are shown in the following figure:

Figure 1: Cross-section of Tail with Variables [1]

In general, silicone is a great binding agent since its material properties allow for high

heat resistance, chemical stability, and electrical insulation. The backbone of silicone is made up

of several silicon and oxygen atoms stringed together which gives it a bond length of 1.63 Å and

a bond angle of 130˚. These longer bond lengths mean that the atoms have more room to move

and are able to change conformation easily. Due to the helical formation of its molecules and low

intermolecular force, it is a very flexible polymer with high elasticity, which is needed for any

deformation to occur. [2]

10 | P a g e

Figure 2: Silicone Backbone [2]

In addition to the Si-O backbone, silicone has a few methyl molecules that are attached to

the outside of the coil structure. These methyl groups are allowed to freely rotate and give the

material high water resistance. Since this tadpole will be tested underwater, these properties will

be essential to the data acquisition. Another property that silicone possess is heat and cold

resistance. Only temperatures below -70C and above 150C will cause the material to deteriorate.

Since these are extreme temperatures, it is safe to say silicone will not fail easily. For future

purposes of implementing a robotic design, silicone retains high electrical insulation which

ranges from a resistance of 1TΩm to 100TΩm. This resistance does not change at all even when

immersed in water.

III. Goal and Objectives

The purpose of this project is to produce a high angle of deformation so that the tail will

create an S-shape similar to the rapid escape response of a tadpole. The desired S-shape can be

achieved with angles between 45 to 80 degrees, whereas a C-shape only needs a 10 to 45 degrees

angle. Testing will be done manually unlike previous methods that involve robotic components

in order to simplify and perfect the core design. A successful outcome would include an angle of

deformation between 45-80 degrees and an apparatus that can feign swimming underwater.

11 | P a g e

IV. Design Process

The process for creating this apparatus consists of two specific parts: design and printing.

The design process will include the head component which will act as a compartment to

transport nitrogen pressure from a tank outside the water into the tail section. The tail section is

the most important aspect of the design process. The channels need to be designed in such a way

that maximizes the deformation angle. Moving on to the printer phase, the program that supports

the printer has certain specifications and settings that have to be adjusted for the purpose of this

project. Learning to use this printer and understanding all of its parameters is just as essential to

the success of this project as the design portion.

A. Printer Specifications

Since this project requires the use of silicone, a material not commonly supported

by most 3D printers, a Wanhao Duplicator 4 printer was ordered and fit to be able to

push out silicone. [3] The basic set-up requires a nitrogen pressure tank which runs to

a tube of silicone. The connecting piece between these two items is removeable at the

mouth of the tube for pressure alleviation once the printing is finished. At the other

end of the tube filled with silicone, a nozzle can be twisted into place. There are

numerous different nozzle sizes that can be interchanged depending on the print job.

For the purposes of this project, a 1.7mm diameter was chosen since the channels do

not need such small, detailed work.

12 | P a g e

Figures 3 & 4: Physical Set-Up for a Wanhao Printer with Connections to

Nitrogen Tanks.

As for the program that runs the printer, ReplicatorG is used to generate a gcode

from an STL file. To generate the code, it needs a slicing profile to follow. Skeinforge

and Slic3r X.0 are already built into the program, but the default settings for these

have to be adjusted for every print job. The types of specifications include: layer

height, infill, speed, retraction length, temperature of the bed, and many more. It takes

many hours of trial and error in order to get the perfect conditions for a specific print

job. If the speed is set too slow, then there will be silicone build up; but too fast will

result in ununiform silicone layers. Before any designing took place, the printer

settings were adjusted based on a solid rectangle around the size of the tail. This was

just a baseline test to make sure everything on the printer would run smoothly. After

testing many different variations of setting codes, the best and smoothest result

occurred using the Slic3r X.0 profile with the following specifications:

13 | P a g e

Figure 5: Print Settings

Figure 6: Printer and Filament Settings

Now that a gcode can be generated with correct settings, an SD card is used to

transfer the gcode to the printer directly. From this stage onward, the printing requires

some manual input. At the control panel, the file has to be chosen from the SD card

which will start the printing process. Part of this process involves heating the bed

plate and center the tube at the bottom left corner of the machine. Once the printer

fulfills these requirements, the tube will start to move to the part’s point of origin. It is

very important to watch the pathway carefully and only turn on the nitrogen pressure

tank when the nozzle reaches that point. Otherwise, the silicone will start to flow

before and possibly create an error in the part being printed. Likewise, once the

14 | P a g e

printing is finished, it is the responsibility of the observer to relieve pressure from the

tube immediately. If this does not happen, silicone will continue to flow and fall on

top of the part. In order to ensure there will be no error in part construction, constant

attention is required. There is an option on the printer to pause or stop the print job if

the observer sees an error in the printing process. Errors can include air bubbles in the

silicone, a low supply of silicone, or just incorrect settings. Knowledge and an

understanding of this unique printer was paramount during research into this thesis.

Although a lot of time was poured into the printer set-up, a fully-functional 3D printer

of silicone could provide many research opportunities for this University in the years

to come.

B. Proposed Designs

There are two very different components that need to be designed: the head of the

tadpole and the tail. The head was the first thing to focus on since it did not require

the use of the silicone printer, and it was a much less complicated design. The shape

was designed to be bulbous and oval like a typical tadpole. However, it will be

hollow on the inside to allow for nitrogen pressure to flow into the tail. For this

reason, a hole was cut from the top of the head to serve as an input for the tube that

runs to the nitrogen pressure tank. Another hole was cut at the back of the head to

connect to the tail portion and top series of channels.

15 | P a g e

Figure 7: Tadpole Head Design.

As for the tail, the first design was created with channels all along the body of the

tail. The top and bottom channels are separated completely with the top portion able

to extrude into the head compartment.

Figure 8: First Design of Tail.

Figure 8 displays the tail cut in half in order to show the channels, but the actual

design would be a closed system with a mirror of the previous image as seen in figure

23 in the appendix. The thought process for this design occurred before the realization

that only a cross-section of the tail needs channels. Also, the testing for this design

happened before a complete understanding of the printer was achieved.

16 | P a g e

Figure 9: Printed Version of First Tail Design.

As seen from figure 9, the first design was a failure. There was no way to position it

on the printer that would not result in total collapse of the air chambers. This is what

led to the idea that a cross-section would be a better route to pursue.

For the second design, some outside help was required. After speaking in detail

with a colleague who has more experience with the Wanhao printer, she advised

designing something similar to her mixer design which also includes air channels. If

the design does not include chamfers that offer an incline between the channel wall

and the outside of the tail, the channels will continue to collapse under the weight of

the silicone. Based on her advice, her mixer design was adjusted to fit the tadpole tail

requirements.

17 | P a g e

Figure 10: Second Design of Tail.

This design allows for better separation between the bottom and top line of channels

and has less probability of failure during printing. Also, it follows the idea that a

cross-section of the tail should be the only part that includes channels. The rest of the

tail will follow the angle of deformation as long as the first section deforms with the

help of the channels. Later, a pure silicone triangle will be designed to stick on the

end of the cross-section, thus forming a complete tail. It is important to note that

when the STL file for this design is inputted into ReplicatorG, the drawing was scaled

up by 1.5. This is true for every part after this as well. As expected, the part came out

much better this time around; but when nitrogen was pumped into the top section, no

deformation occurred. Obviously, nitrogen was getting into the channels, but it only

caused the outer wall to puff up as seen by figure 12.

18 | P a g e

Figure 11 and 12: Printed Version of Second Tail Design.

From further analysis, the problem with this design is that the actual thickness of the

cross-section is too high. When it is too thick, the silicone cannot deform at all after it

is cured. This was proven when physical force could not even deform it. There was

not way that nitrogen would be able to complete the job. The next design will account

for this flaw.

For the third design, the thickness was cut in half to correct the flaw seen in the

last design. Also, the width and height of the channels were increased to make up for

a decrease in number in channels. This decrease needed to occur to keep the cross-

section square.

Figure 13: Third Design of Tail.

19 | P a g e

After printing the previous design on blue tape, there was no way to see the

structure of the channels. The only way I knew that they did not collapse is the

fact that nitrogen pressure could travel through the cross-section without

impedance. This time around, the cross-section was printed on a glass slide.

Figure 14 and 15: Printed Version of Third Tail Design.

Once the silicone was partly cured, the cross-section was peeled off the slide, and

the bottom was sealed with more silicone. After waiting for the part to completely

cure, nitrogen was pumped through the top portion once again to make sure some

deformation will occur. While there was a slight deformation, it was clear that it

was not enough to make a C-shape, much less an S-shape. The next design will

have to be adjusted accordingly.

This next design is very similar to the previous, but the number of

channels were increased from 4 to 8. Also, the shape was changed from a square

to a rectangle. From equation 1 stated in the background research, increasing the

number of channels theoretically should increase the angle of deformation.

20 | P a g e

Figure 16: Final Design of Tail.

This design had to be printed numerous times due to faulty program paths. For

some reason, ReplicatorG can make inefficient pathways by moving the nozzle

from one end of the part to the other. Since the set-up for silicone does not allow

for a stop mechanism when this change occurs, silicone will be dumped the whole

way even in places that the design does not call for silicone. By moving the origin

point and with a little manipulation, a clean part was printed and ready to test.

Already from preliminary testing, a good amount of deformation was noticeable.

Figure 17 and 18: Printed Version of Final Tail Design.

C. Analysis of Final Design

Before any testing can begin, theoretical calculations must be made to

ensure the angle of deformation will be large enough. Due to limitations set by the

nitrogen tank, the pressure range is set from 5 to 30 psi meaning there will be 6

21 | P a g e

data points. Using equations 1-3 and specifications given by the design, a table

was made to find the θ value.

Table 1: Theoretical Values for Theta based on Final Design.

From the table, θ seems to jump between 5 and 10 psi then levels out after values

higher than 20 psi. It seems the working range of deformation occurs between 10

and 20 psi. Also, the max angle of deformation is only approximately 24 degrees.

This is not within the range of S-shape deformation, but it is within the range of

C-shape deformation. Even with manipulation of the number of channels and the

size of channels, the θ value does not increase by much. It would take a drastic

change to the design like doubling the amount of channels for any real change to

take place. Considering the limited amount of time for research, the C-shape

formation was accepted as a valid goal.

D. Testing of Final Design

The goal of testing is to acquire experimental data that is close in value to

the theoretical numbers given in the previous section. To start, a nozzle was

obtained that can fit on the end of the nitrogen pressure tube. This is the same

tube and nitrogen pressure tank that is hooked up to the printer. The other end of

the nozzle is small enough to fit in the hole that leads to the upper channels. By

grabbing on to the nozzle and front corners of the cross-section, nitrogen was able

to be pumped into the tail, and deformation was observed. With the help of

22 | P a g e

another student, videos were taken at each data point and screenshots were taken

at peak angles.

Figure 19: Data taken at 15 psi with a 10 degrees Angle of Deformation.

From figure 19, it is evident that the tail is deforming to create a small C-shape,

but it is not deforming enough to produce a big, clean C-shape which is what was

expected from the theory.

V. Test Results and Discussion

After analyzing all the pictures taken at each pressure point, a table was made to

show the experimental relationship between pressure and θ.

Table 2: Experimental Data for Theta based on Specific Pressure.

A similar trend to the theoretical values occurred where the angle of deformation

increased significantly at 10 psi and then leveled out at 20 psi. Unfortunately, the

experimental values of θ are nowhere near the theoretical values that were expected. In

23 | P a g e

order to compare the two results, a graph was constructed for the angle of deformation

based on pressure.

Figure 20: Graph of Theorical and Experimental Angle of Deformation.

As seen from the graph, the two sets of data follow a polynomial trend line which means

that the theory and formulas for angle of deformation are valid. However, there must

have been several sources of error within the actual design of the cross-section that forced

the experimental data to be so skewed.

The exact percent error difference between experimental and theoretical values

can be observed in table 3.

Table 3: Percent Error between Experimental and Theoretical Values.

24 | P a g e

The percent error ranges from 50% to 100%. These values are way too high to signify a

specific area of error. One possibility of error could come from leakage in the silicone

walls of the cross-section. Leakage was a big problem while printing parts, but any small

holes that developed were patched up with more silicone. Even when testing, there was

no big spout of nitrogen. However, this does not mean there were not any micro-holes in

the walls. This small leakage applied an inaccurate value of pressure through the cross-

section resulting in at least 20% error. One way to fix this would be to make the walls

one more layer thicker. The other way to avoid this problem altogether would be to run

the STL through a program that fixes the holes created from SolidWorks. Another area of

error could result from the assumption of the Young’s modulus value of silicone. In

reality, it could be a lot higher than what was calculated for. The final source of error

could arise from the size of the nozzle that was used on the printer. With a bigger nozzle,

the print time is pretty fast which was important for the time constraint of this project, but

the detail and accuracy of any part decreases significantly. The width and height of the

channels can vary because of this. In future attempts of this project, a smaller nozzle is

recommended to ensure those values are as accurate as possible.

VI. Conclusion

The cross-section tail design performed in a way that emulated a slow swimming

motion of an organic tadpole. Through many hours of printing, a design was crafted that

achieved the acceptable C-shape when nitrogen pressure was pumped throughout the

system. The resulting angles of deformation were much smaller than expected, but they

follow the same trend laid out by the theory. Although this project was not a complete

success, it created a good starting point for future undergraduates at the University of

25 | P a g e

Arkansas to continue the work. This University would benefit a lot from research into

soft robotics. The market is currently exploding with profit and opportunity in this field.

Success in soft robotics research would bring the University to an even playing field with

other scientific colleges.

A consideration that one should have before taking on the responsibility of this

project is the immense time commitment that this requires. Silicone has to cure for at

least 5 hours before any testing can occur. This means months of printing could take

place before a perfect part is produced. Additionally, the learning process to use the

Wanhao printer can take an enormous amount of time. Another recommendation would

be to increase the number of channels and overall size of the cross-section in order to

create bigger angles of deformation. This will be no problem since the continuation of

this subject will deal with a fish instead of a tadpole. Also, the nozzle size will need to be

decreased and the wall thickness increased to ensure no error will take place. Even

though this particular research did not meet the high standard set at the beginning, other

students can learn from these mistakes and create a robotic soft fish that can operate

without fault.

VII. References

[1] Andrew D. Marchese, Cagdas D. Onal, and Daniela Rus. Soft Robotics. Mar 2014.

ahead of printhttp://doi.org/10.1089/soro.2013.0009

[2] “Characteristic Properties of Silicone Rubber Compounds.” Shin-Etsu, Aug. 2016,

www.shinetsusilicone-global.com/catalog/pdf/rubber_e.pdf.

[3] “D4 Series User Manuel.” WANHAO 3D PRINTER, 2012,

www.wanhao3dprinter.com/FAQ/ShowArticle.asp?ArticleID=23.

26 | P a g e

[4] “Materials Selection in Mechanical Design.” Wright State University, 2008,

cecs.wright.edu/~rsrin/Courses/ME474-674/Winter%202008/Slides-9-Full%20Size.pdf.

27 | P a g e

Appendix: Assembly Drawing(s), Exploded View, Component Drawings.

Figure 21: Full SolidWorks Design of Tadpole Head.

Figure 22: Drawing of Tadpole Head.

28 | P a g e

Figure 23: Full SolidWorks Design of First Tadpole Tail.

Figure 24: Drawing of Final Design of Tadpole Tail.

29 | P a g e

Figure 25: Placement of Cross-Section in ReplicatorG.

Figure 26: Example of Problems with Silicone Printer i.e. Channel Collapse.